Komi and winrate statistics

There are statistics from the Japanese games with yearly winrate from this website, that are constantly updated (1st column year, 2nd column yearly matches, 3rd column yearly black wins, 4th column yearly white wins, 5th/final column yearly black win rate)

I’d like to add some statistics I extracted from the hotongo (弘通围棋) of 174,946 professional Chinese internal games from 1926 to 2005

The shared statistic webpage using google sheet

I’ve mentioned before that in 2022, there was a statistic made that year with a black winrate of about 49.7%. And I’ve suspected at the time that as pros kept getting better at using AI as a training/reviewing tool, we might see the effect of a wider winrate gap.

And indeed, we see the gap widening, but not as you would expect. The trend of a slight downward trend from 2019 (51.0%), 2020(49.6%), 2021(49.1%), 2022(49.7%), got reversed from 2023(51.8%), 2024(53.4%), 2025(52.6% so far). The gap is widening, but black winrate climbed back up to over 5.5 komi winrate using Japanese rules (imagined that, Chinese rules 7.5 komi professional games, have higher winrate statistically than using Japanese rules 5.5 komi at this moment). And they are not just some hundreds of games, but thousands, tens of thousands of Chinese games played mostly by Chinese A/B/C league pro players. Their livehood depends on winning these games, or they might get nothing, the training of using AI tools makes Chinese players get better at playing as black as it seems.

7 Likes

I wonder if beyond the statistical calculation, chosing a higher komi (8.5?) will bring some day other difficulties as just a rebalance of points. If there will be a realistic possible strategy to counter the komi.

I feel we need to take a couple of years and recheck the statistic again. When pros started to realize that the “penalty for black” at 7.5 is actually not true (I think a lot of them started to realize that, some pros noticed lot of games are won by black in some of the title matches in all top 4, instead of the couple of years ago more white wins from their perspective). They would again study strategies to counter them as white. It might have happened already this year, and we would see a “return to normal” for a more balanced black winrate next year.

I think even Shin Jinseo also mentioned that taking black is more stressful because you have to prepare a lot, while taking white is like a walk in the park for him. If only we can measure the stress level of those who take white and who take black.

It’s also true that some players win more with white than black?

Some players like Shin jinseo, regardless of how they feel about their stress, would still have extremely high winrate regardless of the color (76% and 79% overall), or like Lee Sedol both around 67~68% as black or as white. There are players like Ke Jie before 2017, had like ~90% win rate as white, but not as high playing as black (still has like 73% overall lifetime winrate as white today, but less than 60% as black). And there are players like the strongest woman player Choi Jeong, who has >70% black winrate because she likes complicated fights and killing big dragons, but very modest win rate playing as white around 60%. And there are players like Li Qincheng (李钦诚) who is probably the “fastest” go pro in the world (in terms of thinking time playing blitz even in “normal time setting”), who had like >66% win rate as black, but just around 50% winrate playing as white, since he relies heavily on momentum, if he lost it he would get lost easily in defense.

4 Likes

You could watch a lot of the Korean Baduk league where they have the heart rate monitors :slight_smile:

1 Like

It will be interesting to see whether there’s a correlation between the playing style and the black winrate. If those who like big fights really win more as black, it also shows that black can’t win by playing normally. :sweat_smile:

1 Like

I’ve watched before but many of them control their heart rate so well it’s hard to tell whether they are stressed or not :sweat_smile:

2 Likes

That’s an interesting question. I just checked players like Mi Yuting (芈昱廷), who is famous for the flying knife/dagger joseki, and loves complicated fight, but his win rate as black/white is pretty close (64%/63%). Or Ichiriki who is also more towards the fighting style, but has black/white winrate (65%/~70%) slightly better as white. Also, players like Lai Junfu (賴均輔) who is considered as the strongest fighting style player in Taiwan, but he has like 64% winrate as white, but not even on the list for players have high black winrate (meaning like just around 50%), which is actually quite shocking to me, but at the same time not, since he has a “counter-attack” style, often find very tricky play when he is behind, so not that kind of “fight and kill” using strength, but throw-in, trick play and then big semeai fight at the late mid-game (and his early mid-game fight, failed more often than not)

1 Like

Here are some updates from the recently held International tournaments using Chinese rules with 7.5 komi.

2025 Yang-ming “Polarbear” Cup (invitation tournament) (if anyone wonders about the name, Yang-ming was the name of a Chinese Ming Dynasty Confucius school, and Polarbear was the sponsor’s brand name)
{B wins}: 9
{W wins}: 6
black winrate 60.00%

The 3rd Quzhou-Lanke Cup
{B wins}: 29
{W wins}: 20
black winrate 59.18%

Also, for the 15th National Sports Tournaments for the Weiqi event for the general public held in Sep/Oct (第十五届全国运动会群众比赛围棋项目, China internal national tournaments, but for all players, amateurs included, competed alongside pros)
{B wins}: 19
{W wins}: 13
black winrate 59.38% for personal tournaments only. But if we include team games, and preliminaries (from the records I have), the results are:
{B wins}: 76
{W wins}: 55
black winrate 58.02% (not much of a difference)

However, for the International tournament the 1st Beihai Xinyi Cup held earlier this year in April
{B wins}: 27
{W wins}: 37
black winrate 42.19%

And I double checked whether I accidentally mixed black with white, and they really have such a huge discrepancy. Although they would average out to 65 {B wins} and 63 {W wins}, with black winrate 50.78% for International games only, and black winrate 54.44% including the 15th Chinese National tournament records.

And the craziest statistic is the Chinese A league (the 27th A league), which is still ongoing and less than halfway done. Currently, the winrate for black is 43.70% from 254 matches, while for the Chinese C league, the figures are on the level of close to 54% from samples of a quadruple number more games records (there are 30 teams for the C leagues, but only 16 teams for the A league, and the records for C league are incomplete, since people have low interests in watch them, but black definitely have much higher winrate from the samples I saw)

However, the winrate from the A league in 2024 (a few of the final 26th A league games were played in January this year, and not counted) was 48.09% over 576 records (277 black wins, 299 white wins). The higher black winrate from 2024 almost exclusively came from lower-tier B and C league games. However, the 2025 records were still very odd compared to last year, with exceptionally uneven winrate across different tournaments using the Chinese rules. And from the current records so far, I’d predict black winrate in 2025 for pros would still favor black for quite a margin (higher or lower than 2024 hard to tell, and need to consider if we include national tournaments that include amateurs, since they have exceptionally high black winrates)

And the only trend I can sort of justify this discrepancy and gaps is that at the low-end of the pro to strong amateurs, players seem to have higher winrates playing as black, but it dropped sharply on the average pro levels to almost even, and likely still favors white for the top pros (but even top pros are getting better at playing as black, like the 3 finals for the Quzhou-Lanke Cup were all won by black). And it seems to be rising across all regions, not just for Chinese players.

3 Likes